Motion 4: P1788 on non-754?
Friends,
I do not have better wording to suggest, but here are some thoughts on the
issue.
At one level, how something may be implemented is not the concern of the
standard. At that level, the underlying hardware is irrelevant. At another
level, if something is impossible to implement (efficiently, depending on
one's point of view), there is no point putting it into a standard. Hence,
discussions leading to a standard SHOULD sometimes include discussions about
implementations. That is messy enough if the discussions are about possible
implementations in 754. The discussion complexity is probably unbounded if
we admit discussions of possible implementations on ANY architecture, known
or imagined.
I conjecture that part of John's motivation may have been to avoid at least
some of the discussions of IA implementation on MANY different
architectures.
Out best hope of getting a standard enacted is to Keep It Simple, as least
as simple as possible. I want to say, "This is about IA on 754 hardware.
That is ambitious enough. We'd LOVE it if you want to do IA on non-754
hardware. If you do, we hope you'll make it look as much like what we said
in P1788 about 754 as you can. P1788 may be a useful guide, but it does not
apply as a standard.
The drive for this standard is coming primarily from academics, most of long
standing in the IA world. Some of us may see this as a chance to validate
our life's work. Sorry, but that is a pretty poor motivation for a
standard. Especially since there is precious little commercial push, the
more like a Christmas tree (hang everything on it) we make it, the less
likely it is to be approved, and if approved, commercially implemented, and
if commercially implemented, widely used.
I'd LOVE to see IA in every application, in every language. I'd love to see
IA in fast hardware, as Prof. Kulisch has championed for years. I'd love to
see all the richness and expressiveness of csets, modals, and any of the
other variations our community has developed during the past 50 years.
However, we have not made much progress on wide-spread acceptance in 50
years. Rather than craft a standard that standardizes all our work, I think
a standard that codifies the CORE stands a better chance of actually
bringing IA to the masses. We cannot force on the rest of the world our
view of how scientific computation SHOULD be done; the best we can do is
offer as easy to understand and as easy to use a platform for IA and hope
for the best.
Who knows? At one point, Frontline Systems showed a version of Excel that
used intervals behind the scenes to help ensure correct results.
I'm flexible on the details, but let's Keep It Simple. Any language lawyers
know how to state that?
Dr. George F. Corliss
Electrical & Computer Engineering
Haggerty Engineering #296
Marquette University
P.O. Box 1881
1515 W. Wisconsin Ave.
Milwaukee WI 53201-1881
George.Corliss@xxxxxxxxxxxxx
414-288-6599; -288-4400 (GasDay); -288-5579 (Fax)
Www.eng.mu.edu/corlissg